Quaternary Science Journal GEOZON SCIENCE MEDIA Volume 63 / Number 1 / 2014 / 44-59 / DOI 10.3285/eg.63.1.03 ISSN 0424-7116 E&G www.quaternary-science.net Ice-Rafted Erratics and Bergmounds from Pleistocene Outburst Floods, Rattlesnake Mountain, Washington, USA

Bruce N. Bjornstad

How to cite: Bjornstad, B. N. (2014): Ice-Rafted Erratics and Bergmounds from Pleistocene Outburst Floods, Rattlesnake Mountain, Washing- ton, USA. – E&G Quaternary Science Journal, 63 (1): 44–59. DOI: 10.3285/eg.63.1.03

Abstract: Exotic ice-rafted debris from the breakup of ice-dammed glacial lakes Missoula and Columbia is common in slackwater areas along the 1,100-km route for outburst floods in the northwestern US. A detailed analysis was performed at Rattlesnake Mountain, which lay beyond the limit of the former ice sheet, where an exceptionally high concentration of ice-rafted debris exists midway along the floods’ path. Here floodwaters temporarily rose to 380 m elevation (forming short-lived Lake Lewis) behind the first substantial hydraulic constriction for the outburst floods near Wallula Gap. Within the 60 km2 study area more than 2,100 erratic isolates and clusters, as well as bergmounds were recorded. Three quarters of erratic boulders are of an exotic granitic composition, which stand in stark contrast to dark Columbia River basalt, the sole bedrock in the region. Other exotics include Proterozoic quartzite and argillite as well as gneiss, diorite, schist and gabbro, all once in direct contact with the Cordilleran Ice Sheet to the north. Most ice-rafted debris is concentrated between 200 and 300 m elevation. Far fewer erratics and bergmounds lie above 300 m elevation because of the preponderance of less-than-maximum floods. Plus, larger deep-rooted icebergs were forced to ground farther away from the ancient shorelines of transient Lake Lewis. As floodwaters moved across the uneven surface of Rattlesnake Mountain, many erratic-bearing icebergs congregated into pre-existing gullies that trend crosswise to flood flow.

Eisverfrachtete Findlinge und Bergmounds aus Ausbruchsflutwellen im Pleistozän,R attlesnake Mountain, Washington, USA

Kurzfassung: Eisverfrachteter Schutt findet sich häufig im Stauwasserbereich der 1.100 km weit reichenden gigantischen Ausbruchsflutwellen aus den eiszeitlichen Missoula- und Columbia-Seen im Nordwesten der USA. Eine detaillierte Analyse erfolgte am Rattlesnake Moun- tain. Dort ist eine außergewöhnlich hohe Konzentration dieses Schutts in der Mitte des Gerinnebettbodens zu finden. Zeitweilig bildete der Flaschenhals des Wallula Gap die erste wesentliche hydraulische Verengung und ließ die Gletscherflut vorübergehend auf 380 m Seehöhe ansteigen, wodurch kurzzeitig der Lewis See gebildet wurde. Auf einer Fläche von 60 km2 wurden mehr als 2.100 Findlinge, Ansammlungen von erratischen Felsbrocken und Bergmounds registriert. Im Gegensatz zum lokal vorkommenden dunklen Columbia River Basalt bestehen drei Viertel der Findlinge aus granitartigem Material. Auch Schutt von dem ehemals im Norden verlaufenden Eisgebirgszug wie sedimentärer Quarzit aus dem Proterozoikum, Tonschiefer, Gneisgestein, Diorit, Schiefer und Paulitfels wurden hier gefunden. Ein Großteil des eis-verfrachteten Schutts befindet sich auf 200–300 m Seehöhe. Weit weniger Findlinge und Bergmounds sind über 300 m Seehöhe anzutreffen, da es überwiegend Fluten von sub-maximalen Ausmaßen gab. Außerdem liefen größere Eisberge aufgrund ihres Tiefgangs weit von der Küstenlinie des vorübergehend bestehenden Lake Lewis auf Grund. Bei der Flutbewegung über den unebenen Untergrund des Rattlesnake Mountain blieben viele Findlinge in bereits vorhandenen, quer zur Flutrichtung verlaufenden Wasserrinnen hängen.

Keywords: ice-rafted debris, erratic, bergmound, Missoula floods, Wallula Gap, Lake Lewis, glacial Lake Missoula, Wisconsin Glaciation, Columbia River basalt

Address of author: Bruce N. Bjornstad, P.O. Box 999, MSIN K6-81, Pacific Northwest National Laboratory, Richland, Washington 99354. E-Mail: [email protected]

1 Introduction and Background Ice Sheet (Bretz et al. 1956; Bretz 1969; Baker 1978; Baker & Bunker 1985; Allen et al. 2009; Smith 1993; Waitt 1980, Characteristics and distribution patterns of ice-rafted de- 1985). Other sources for Ice Age floods in the area include bris (erratics and bergmounds) can provide insight into the at least one flood from ice-dammed glacial Lake Columbia flow dynamics as well as possible source(s), timing, and (Atwater 1987; Waitt 1994; Waitt et al. 2009), Lake Bon- frequency of Ice Age flood events. Erratics in the Pacific neville (O’Connor 1993) and possibly from one or more Northwest have long been recognized in areas downstream subglacial outbursts from beneath the Cordilleran Ice Sheet of the maximum extent of glacial ice (Bretz 1919, 1923a, itself (Shaw et al. 1999). The Cordilleran Ice Sheet never 1923b, 1930, 1969; Allison 1933, 1935; Bretz et al. 1956; advanced south into the mid-Columbia Basin where much Fecht & Tallman 1978; Minervini et al. 2003). Multiple ice-rafted debris came to rest. Therefore, the only plausible cataclysmic outburst floods, mostly from ice-dammed gla- explanation for high-elevation erratic debris beyond the ice cial Lake Missoula (Figure 1), are now generally accepted as front is from floating icebergs carried during outburst-flood the source for erratics beyond the limits of the Cordilleran events.

44 E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 120o 118o 116o 114o 49o Cordilleran Ice Sheet

Columbia Lobe Colville Okanogan Lobe Lobe o 48 Glacial Lake Columbia Ice Dam o u n t a i n s

M Spokane R o c k y Wenatchee Glacial Lake Missoula

47o Vantage M o u n at ni s Missoula

C a s c a d e Yakima Study Area

o WA N 46 Fig. 2 ID OR o u n t a i n s Columbia R. Gorge Wallula M Gap ID MT

B l u e

0 100 mi. Pleistocene lake Higher-Energy Flood Deposits 0 100 km

Palouse loess (>1m thick) Slackwater Flood Deposits

Fig. 1: Ice Age glacial and flood features over a portion of the northwestern United States. Abb. 1: Eigenschaften von Gletschern und Gletscherläufen im Nordwesten der USA.

Although the largest Missoula floods drained within sev- bedrock-constrained reaches between basins like Wallula eral days, up to three weeks were required for the flood- Gap and the Columbia River Gorge (Figure 1). waters to completely pass through several hydraulic con- The earliest floods occurred during one of many previ- strictions along route (Waitt et al. 2009; Denlinger & ous glacial cycles >780,000 yr ago in the early Pleistocene O’Connell 2010). The first major constriction for the out- (Patton & Baker 1978; Bjornstad et al. 2001; Pluhar et al. burst floods was at Wallula Gap where, after spreading out 2006; Bjornstad 2006). The last period of flooding occurred across a 160 km-wide tract of the Channeled Scabland, flood- during the Late Wisconsin Glaciation between 20,000– waters were forced through a single, narrow opening only 3 15,000 cal years BP (O’Connor & Benito 2009) when as km wide (Bjornstad et al. 2007). During the largest floods many as 100 discrete flood events may have occurred At( - Wallula Gap and the Columbia River Gorge downstream water 1986; Waitt et al. 2009). Many of these floods were could transmit only a portion (10 +/– 2.5 million m3/sec) of proportionally small (a few million m3/sec) relative to dis- all the floodwater entering the Pasco Basin O’Connor( & charge for the largest (>17.5 million m3/sec) Late Wisconsin Baker 1992; Benito & O’Connor 2003). Subsequently, a outburst flood(s) from glacial Lake Missoula O’Connor( & huge temporary lake (Lake Lewis shown in Figure 2) backed Baker 1992). up to a maximum elevation of ~380 m (1,250 feet) behind this The majority of erratics, which consist of light-colored bedrock constriction (Bretz 1923a). Ice-rafted debris float- plutonic and metamorphic rocks, stand out in sharp con- ing in Lake Lewis was sequestered in slackwater areasalong trast to the dark, Miocene, Columbia River Basalt Group – basin margins and backflooded valleys(Figures 2 and 3). The the only bedrock native to southeastern Washington State stranded icebergs eventually melted, forever leaving behind (Figure 6). Most ice-rafted debris appears derived from the their payloads of exotic detritus (Figures 4 and 5). breakup of the Purcell Trench Lobe of the Cordilleran Ice Maximum heights for the floods are indicated by strand- Sheet that temporarily dammed glacial Lake Missoula (Fig- lines, scarped hills, ice-rafted erratics, and divide crossings ure 1). Ice-rafted debris may also be associated with the final (Baker 1978). The upper limit of the ice-rafted debris de- breakup of the Okanogan Lobe that blocked glacial Lake Co- scends downstream in a stair-step fashion as floodwaters lumbia until the end of the Wisconsin Glaciation. passed through a series of constrictions en route (Baker The detailed study of the nature and distribution of ice- 1978; Benito & O’Connor 2003). Water-surface profiles of rafted erratics and bergmounds provides valuable data sets floods were relatively flat within basins but steepened in to test various hypotheses regarding the history and rela-

E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 45 Fig. 3 Othello

Yakima

Rattlesnak M ount ai n e

N Richland

Wallula Walla Gap Walla

Fig. 2: Maximum extent (up to 380 m elevation) of back flooding (Lake Lewis in blue) behind hydraulic constriction at Wallula Gap during the largest outburst flood(s). Abb. 2: Maximale Ausdehnung der Rückstauflut (bis zu 380 m Seehöhe) hinter der hydraulischen Verengung des Wallula Gap (blaue Markierung: Lewis- See).

120o o o 119 30' 119 tive sizes of Ice Age floods. Only a few previous studies ex- Slackwater Area ist for erratics emplaced by outburst floods. These include a study by Allison (1935) within the Portland Basin and later expanded upon by Minervini et al. (2003). Another study 46o40' by Karlson (2006) was performed in the Ginkgo Petrified Forest State Park, located 80 km northwest of Rattlesnake Mountain near Vantage. The present study gathered an ex- haustive set of data including lithology, size, roundness, 46o30' shape, and weathering characteristics of more than 2,000 in- dividual erratics, along with their locations and elevations Rattlesnake (Bjornstad et al. 2007).

Mtn. 46o20' 2 Study Area

Richland A comprehensive analysis of ice-rafted debris was perfor- med in a 60 km2 area along the north flank of Rattlesnake 46o10' Mountain – the tallest segment of an elongated northwest- southeast trending anticlinal ridge (Yakima Fold) of Colum- N bia River basalt (Figure 6) that rises to 1,075 m elevation. 0 10 mi Wallula Gap During the largest outburst floods, water rose to within 700 0 10 km m of the summit, making Rattlesnake Mountain a long pe- ninsular land body during flooding (Figure 2). The study ar- Fig. 3: Shaded-relief map showing variable flow of floodwaters through the Pasco Basin. White areas were above maximum flood level (>380 m ea is located along the low-relief northeastern flank of the elevation). Sizes of arrows are approximately proportional to flow velocity. ridge where floodwater flowed from the northwest to the Stippled pattern signifies slackwater areas that received the greatest quan- southeast (Figure 3). The upper third of the study area con- tities of ice-rafted debris. The Rattlesnake Mountain study area is located sists of a gentle (~2o), northeast-dipping surface (Iowa Flats) at center. that reflects the surface on the underlying basalt bedrock. Abb. 3: Schattierte Reliefkarte zur Darstellung der Flussvariablen der Only a few metres of relief exist within small gullies across Fluten im Pasco-Becken. Weiße Bereiche zeigen maximalen Wasserstand Iowa Flats. Downslope of Iowa Flats ephemeral streams are (> 380 m Seehöhe). Das Pfeilmaß ändert sich proportional zur Strömungs- more entrenched into the basalt displaying as much as25 geschwindigkeit. Diagonale Linien bezeichnen Stauwasserbereiche mit den größten Mengen von eisverfrachtetem Schutt. Das Untersuchungsgebiet m relief between the tops and bottoms of gullies. Although Rattlesnake Mountain befindet sich im Zentrum. basalt may be exposed at the bottoms of gullies, inter-fluvial areas are covered with up to several metres of mostly mas- sive, fine-grained deposits of Quaternary-age loess, slope- wash, or slackwater-flood sediment.

46 E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License Most of the study area lies within a long-protected ecol- about 1,250 feet [380 m] above tide.” Bretz and others, how- ogy preserve, now part of the Hanford Reach National ever, categorized ice-rafted debris into two classes: erratics Monument managed by the United States Fish and Wildlife or bergmounds. The present study distinguishes an impor- Service (Bjornstad 2006; Bjornstad et al. 2007). Closed to tant third category of ice-rafted debris – erratic clusters. Ex- the public for the last 60 yr, this area is ideally suited to the amples of the three types of ice-rafted debris are shown in study of ice-rafted debris because it has escaped widespread Figure 4. Isolated erratics (Figure 4A) generally consist of a anthropogenic disturbances common to other parts of the solitary boulder while clusters (Figure 4B and 5) represent a region. Furthermore, light-colored erratic boulders stand out close grouping of several or more erratics. Bergmounds (Fig- in stark contrast to indigenous black Columbia River basalt ure 4C) consist of distinct, low-relief mounds covered with and the pale-brown blanket of fine-grained Quaternary sedi- ice-rafted debris, up to 30 m or more in diameter and several ments. These attributes, along with the low, sparse, shrub- meters tall (Bretz 1930; Allison 1933; Fecht & Tallman steppe vegetation facilitate the identification and mapping of 1978; Chamness 1993). In general, bergmounds are distin- ice-rafted debris. Last, the gentle yet uneven slopes provide guished from erratic clusters by the presence of a conical an opportunity to examine subtle changes in flood-flow dy- mound displaying recognizable topographic relief. Individu- namics reflected in the distribution of ice-rafted debris. al clusters and bergmounds may contain a single rock type, but more are represented by multiple lithologies (Allison 3 Ice-Rafted Debris 1935; Fecht & Tallman 1978). The areal distribution for all three different types of ice-rafted debris is shown over a por- Bretz (1923a: 605) accurately noted: “A widespread sub- tion of Rattlesnake Mountain on Figure 7. mergence of the lower Columbia Valley is known to have occurred during the Wisconsin glaciation. It is recorded by 4 Methods berg-floated erratic boulders, some of great size, scattered widely in the Columbia Valley below the present altitude of The 60-km2 Rattlesnake Mountain study area was field map-

Fig. 4: Types of ice-rafted debris on Rattlesnake Mountain. (A) Isolated granodiorite erratic boulder at (183 m elevation) in basalt-floored gully. (B) Cluster of granodiorite erratics at 335 m elevation. (C) Pair of bergmounds (~245 m elevation); notice granitic erratic atop left bergmound (encircled). Looking north into the expansive Pasco Basin. Abb. 4: Typen von eisverfrachtetem Schutt am Rattlesnake Mountain. (A) Vereinzelte erratische Gesteinsblöcke aus Granodiorit in basaltgrundiger Schlucht- rinne (auf 183 m Seehöhe). (B) Anhäufung von erratischen Gesteinsblöcken aus Granodiorit Findlingen (auf 335 m Seehöhe). (C) Pärchen von Bergmounds (~245 m Seehöhe); man beachte den Granitfindling (eingekreist) oben auf dem linken Hügel. Blick nach Norden in das weitläufige Pasco-Becken.

E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 47 ped in a series of roughly parallel to diagonal, up-down transects. The transects traversed gullies as well as the inter- fluvial areas, usually spaced 100 m or more apart. All ob- served ice-rafted debris >0.093 2m maximum exposed cross sectional area was located using a hand-held GPS receiver unit. An arbitrary cut-off value of 0.093 m2 was chosen be- cause that was near the minimum size readily observed in the field. Depending on location, erratics can be partially to almost totally buried by Quaternary-age fine-grained loess, slopewash, or slackwater flood deposits. For this reason it was not possible to determine the true volume of most er- ratics and the size reported here represents only the maxi- mum exposed cross-sectional area observed in the field. For data manageability, because many erratic clusters and berg- mounds contain dozens or more individual erratics, specific Fig. 5: Cluster of moderately weathered, banded argillite boulders at 244 m information was recorded only for the largest erratic in each elevation on Iowa Flats along the north flank of Rattlesnake Mountain. The cluster or bergmound grouping. largest erratic boulder is about three metres long. Argillite is derived from Horizontal accuracy of handheld GPS units is generally the Proterozoic Belt Supergroup once in contact with the ice dam for gla- cial Lake Missoula in northern Idaho ~650 km to the northeast. Note that within 3 m while the measured vertical accuracy is much the erratics in this cluster are partially buried in an unknown quantity of less precise. For this reason, instead of the elevation record- Quaternary loess, slopewash and slackwater flood deposits. ed by the GPS, elevations were obtained from a topographic Abb. 5: Haufen von mäßig verwitterten Gesteinsbrocken aus gestreiftem map after location coordinates were entered into National Argilit (auf 244 m Seehöhe) auf den Iowa-Untiefen entlang der Nord- Geographic’s TOPO!© program. Because there is wide spac- Flanke des Rattlesnake Mountain. Der größte Findling links der Bildmitte ing of contour lines in the generally low relief of the study ist etwa drei Meter lang. Argilit (Tonschiefer) entstammt dem Gürtel der area, recorded elevations are believed to be accurate to with- Übergruppe aus dem Proterozoikum, der einst mit dem Eisdamm des ca. in at least half of the contour interval (12.2 m), or about 6 m 650 km nordöstlich entfernten Gletschers von Lake Missoula in Nord-Idaho in Verbindung stand. Man beachte, dass die Findlingsblöcke dieser Gruppe vertical distance. teilweise unter unbestimmten Mengen von Löss-, Ausspülungs- und Stau- In addition to location and elevation, data were collected wasser-Ablagerungen aus dem Quartär begraben sind. on lithology, roundness, shape, and surface characteristics, such as degree of weathering, wind polish, and striations. Because rates of weathering and roundness vary with lithol- ogy, these parameters were compared for only the dominant rock type (i.e., granitic). Granitic erratics classified asun-

o 120o 118o 116 49o Time Rock!Units

Columbia!River!Basalt!Group!(volcanic) Cenozoic Plutonic Ma Sedimentary 66 Mesozoic Ultramafic 48o !

250 and Ice!dam!for !

Glacial!Lake volcanic Missoula Paleozoic metamorphic High-grade 550 Metasedimentary 47o Proterozoic (Belt!Supergroup)!

Yakima!Fold!anticline

WA ID Cordilleran!Ice!Sheet Study!area 46o *

Fig. 6: Distribution of major rock units in eastern Washington and northern Idaho. Most ice-rafted erratics found on Rattlesnake Mountain can be traced to granitic (plutonic) and Belt Supergroup metasedimentary rocks once in contact with the Cordilleran Ice Sheet north and east of the Columbia Plateau. To the south, Yakima Folds strongly influenced the flow of floodwater into and out of the Pasco Basin. Abb. 6: Verteilung größerer Felseinheiten im östlichen Washington und Nord-Idaho. Der Ursprung der meisten eisverfrachteten Findlingen auf Rattle- snake Mountain kann zu den granitartigen (plutonischen) und metasedimentären Felsen des Gürtels der Übergruppe zurückverfolgt werden, die einst in Verbindung mit dem Cordillerischen Eisschild nördlich und östlich des Columbia Plateau standen. Im Süden hatten die Falten der Yakima Senke starken Einfluß auf den Fluss der Flutwasser in das Pasco-Becken hinein und hinaus.

48 E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License Ice-Rafted Debris Bergmound Erratic cluster Isolated erratic Above maximum flood (>380 m elev.)

Cosmogenic-exposure age date

R a tle s n a k e R id g Fig. 7: Distribution of the three different types of ice-rafted debris across the study area. Details on the three erratics sampled e for cosmogenic-nuclide exposure dating

r are listed in Table 1.

Rive Abb. 7: Drei verschiedene Typen von eis- e verfrachteten Schutt sind ungleichmäßig Firing Rang N im Untersuchungsgebiet verteilt. Einzelin-

formationen zu drei Findlingen, die für die Radionuklidmessung des Betrahlung- kima salters ausgewählt wurden in Tabelle 1 Ya aufgelistet. weathered are bright white and show no surficial oxidation boulders that were: 1) well beyond the limits where rocks or mineral decomposition. In contrast, strongly weathered could have rolled or moved down from higher elevations via granitic erratics displayed advanced stages of mineral de- mass movement, and (2) had several metres or more of fine- composition and surface roughness, often with deep red- textured post-basalt sediment between basalt boulders and dish brown or dark varnished surfaces. Slightly weathered underlying bedrock were classified as erratics. to moderately weathered erratics displayed intermediate stages of roughness, oxidation and mineral decomposition 5 Results and Observations (Figure 8). Cosmogenic 36Cl exposure-age dates were obtained from On Rattlesnake Mountain, isolated erratics (52% of all ice- three large, dispersed, ice-rafted erratics reported in Kesz- rafted debris) far outnumber erratic clusters (29%), which thelyi et al. (2009) and represented in Figure 7 and Table 1. exceed bergmounds (19%). However, because many errat- Two of the analyses were performed on granitic boulders; ics smaller than a metre may be concealed beneath a cover the third was performed on an ice-rafted boulder of basalt. of vegetation, the true proportions of isolated erratics and Because basalt is the only indigenous rock type and com- clusters may be greater than that indicated above. The high- pletely underlies the study area and crops out in the steep est concentration of ice-rafted debris lies near 210 m with a slopes of Rattlesnake Mountain to the southwest, special care general decrease above and below this elevation (Figure 9). is needed to positively identify basalt erratics. Some might As previously noted by Bretz (1930) bergmounds are mostly argue that basalt does not qualify as a true “erratic” since it concentrated below 260 m elevation, and generally absent makes up the underlying bedrock at Rattlesnake Mountain, above 300 m (Figures 7 and 9). however, in this paper basalt is included as an erratic when As discussed above, mapping of erratics on Rattlesnake it clearly has an ice-rafted origin. Accordingly, only basalt Mountain was limited to only those with an exposed cross-

E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 49 Fig. 8: Stages of weathering generally indicated by degree of surface roughness and amount of iron oxidation. All erratics are granitic except for the dior- ite boulder at upper left (A). Note advanced stage of surface roughness in photo E whereby the more-stable feldspar phenocrysts stand out in relief due to differential weathering of mineral grains. Abb. 8: Verwitterungsphasen werden im Allgemeinen durch den Grad der Oberflächenrauhheit und den Umfang der Eisen-Oxidation gekennzeichnet. Alle Findlinge sind Granite – mit Ausnahme des Diorit-Findlingsbrockens oben links (A). Man beachte das fortgeschrittene Stadium der Oberflächenrauhheit in Foto E, Aufgrund unterschiedlichen Witterungsverhaltens der Mineralbestandteile treten die stabileren Feldspat-Phänokristalle reliefartig hervor. sectional area >0.093 m2. The majority (60%) were <0.37 m2, 12). The majority (~75%) are light-colored granitic, crystal- while less than 1% were larger than 6.0 m2 (Figure 10A). The line rocks (Figures 10B) like those shown in Figures 4A, 4B, largest erratic (12.5 m2) found in the study area occurred as and 8. However, the number of mapped erratics may be bi- an isolate at a relatively high elevation (310 m). ased in favor of light-colored granitic rocks because they The distribution of erratics by lithology is shown in Fig- are more visible in the field. Most granitic erratics consist ure 11. More than 95% of erratics consist of rocks that are of granodiorite but also include true granites. Granitic rocks very different from indigenous, dark-colored basalt (Figure are 10 times more abundant than the next most common

m ft 1200 n = 2112 350

1100

1000 300

900 Erratics 250 800 Erratic clusters Bergmounds Elevation

700

200

600

500 150 Fig 9: Concentration of ice-rafted debris vs. elevation. Number Abb. 9: Konzentration von eisverfrachtetem Schutts in Bezug zur Höhenlage.

50 E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 682 80 1636

30 584 60

441 20 % 40 % 255 10 20

76 155 112 24 89 73 8 25 21 7 14 0 0 granitic quartzite diorite argillite basalt gneiss schist gabbro other 1-2 2-4 4-8 8-16 16-32 32-64 > 64 ft2 0.09-0.19 0.19-0.37 0.37-0.74 0.74-1.49 1.49-3.00 3.00-5.95 >5.95 m2 Size Class A (maximum cross-sectional area) B Lithology

60 882 824 40 689

649 40 % % 353 20 20 282 301 196

57 7 moderate to slight to angular to subangular to 0 unweathered strong strong 0 rounded moderate angular subangular subrounded

C Weathering D Roundness

Fig. 10: Frequency distributions of Rattlesnake Mountain erratics by: (A) size, (B) lithology, (C) weathering, and (D) roundness. Weathering and roundness pertain only to erratics of granitic composition for reasons explained in the text. Abb. 10: Dichte der Findlingsverteilung am Rattlesnake Mountain unterteilt in: (A) Größe, (B) Lithologie, (C) Verwitterung und (D) Rundheit. Verwitte- rungsgrad und Rundheit beziehen sich ausschließlich auf granitartige Findlinge, (Begründung im Text). rock type within the study area – quartzite. Quartzites oc- decomposition, and oxidation rinds. Many strongly weath- cur in a wide variety of colors, including pastel shades of ered granitic erratics show either a reddish-brown oxidative blue, green, yellow, pink, and orange as well as white and coating or occasionally a spalled, high-relief surface (Fig- gray. Diorite, an intrusive igneous rock containing about ure 8E). Because different lithologies weather at different equal amounts of speckled light and dark minerals, is the rates, a comparison of weathering reported here was only third most common rock type. Next most abundant is argil- performed on only the most common rock type (granitic), lite, an extremely hard, siliceous, dull gray to reddish brown which makes up three quarters of all erratics (Figure 10B). rock that weathers and parts evenly along distinctive bed- More than 80% of the granitic erratics appear moderately ding planes (Figures 5 and 12). The fifth most common errat- to strongly weathered (Figure 10C). Striations and grooves, ic type is Columbia River basalt. Other, less-common erratic presumably derived from glacial abrasion at their places of lithologies include gneiss, schist, and gabbro. origin, were also occasionally observed on surfaces of ice- The most notable surface characteristic on erratics is de- rafted boulders. gree of weathering, which includes wind polish, mineral The surface roundness of erratic clasts varies from angu-

Table 1: Characteristics and 36Cl-exposure ages for three, scattered, ice-rafted erratics on Rattlesnake Mountain. See Figure 7 for relative locations within the study area. Tabelle 1: Charakteristika und Ergebnisse der 36 Cl- Altersbestimmung von drei verstreut liegenden, eisverfrachteten Findlingen am Rattlesnake Moun- tain. Jeweilige Position im Untersuchungsgebiet: siehe Abbildung 7.

Field IDSam ple Cl-36 Latitude Longitude Elevation M axim um Roundness Lithology Shape W eathering No.* Age* (north) (w est) in m (ft) Exposed Area in m 2 (ft2)

E-223 MF-6 16,930 46.37878 -119.50037 310 (1,016) 12.5 (135) subrounded granite elongate strong

E-276 M F-7 16,170 46.38400 -119.46353 209 (686) 5.1 (55) rounded granite equant strong

MERYK1 MF-8 16,740 46.38093 -119.46473 214 (702) 1.1 (12) subrounded basalt irregular m oderate

* from Keszthelyi et al. (2009)

E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 51 0 1 2 3 km Rock Type Granitic Basalt 0 1 mi Quartzite Gneiss Diorite Schist Argillite X Gabbro o 46 26' Above maximum flood (>380 m elev.)

Rattlesnake Mountain Study Area

o 46 24'

o 46 22'

R eal ts n a k eR di g e

River a kim Firing RangYa N e Fig. 11: Distribution of erratics by lithology.

o Abb. 11: Verteilung von Findlingen gemäss 46 20' Lithologie. o o o o 119 32' 119 30' 119 28' 119 26' lar to rounded (Figure 10D). However, the majority of errat- that lies between 200 to 300 m elevation (Figures 7 and 9). ics were classified as subangular to subrounded. Shapes of Fewer erratics and very few bergmounds lie above 300 angular to subangular boulders, argillites for example, often m, despite the fact that the largest floods rose up to 80 m part along parallel bedding planes, leaving them flat to tabu- higher. lar, or rectangular. Plutonic crystalline rocks, on the other hand, are overall more equi-dimensional to elongate, often 3.) Many isolated erratics and erratic clusters are aligned forming a pyramid shape. In general, the shape of over one- with gullies (Figure 7). At higher elevations erratics lie third of all the erratics is irregular. at the bottoms of low-relief gullies. In contrast, at lower Cosmogenic-exposure ages on three Rattlesnake Moun- elevations erratics are usually perched onto the north (up- tain erratics (Table 1) fall into a relatively narrow range be- current) sides of the higher-relief gullies. tween 16,000 to 17,000 cal yrs BP (Keszthelyi et al. 2009). These ages fall within the range (15,000 to 20,000 calendar 4.) A higher concentration of erratics lies along roadways. years) for Late Wisconsin Missoula floods reported else- where (O’Connor & Benito 2009). 5.) Lithology of erratics versus elevation appears random The distribution patterns of ice-rafted debris offer impor- (Figure 11), as does the distribution of erratic size (Figure tant clues to the way the debris was deposited. Summarized 13). here: 6 Discussion 1.) The highest observed erratic found within the study area lies at an elevation of ~362 m. Concentrations of Ice-Rafted Debris Different processes are likely responsible for erratics, erratic 2.) Most ice-rafted debris within the study area is distributed clusters and bergmounds. Isolated erratic boulders repre- along a broad band at the base of Rattlesnake Mountain sent either dropstones from free-floating icebergs, or from

52 E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License Fig. 12: Erratic lithologies on Rattlesnake Mountain. Abb. 12: Lithologie der Findlinge vom Rattlesnake Mountain. smaller, relatively “clean”, grounded icebergs. Erratic clus- the basin like Rattlesnake Mountain and backflooded valleys ters likely melted out of small to moderate-sized, grounded such as the Yakima and Walla Walla valleys (Figures 1 and icebergs, while bergmounds are probably associated with 2), especially during the waxing stages of flooding (Baker especially large, sediment-ladened and grounded icebergs et al. 1991). Far fewer erratics occur in unrestricted, higher- (Bretz 1930). gradient reaches of the floods, like Wallula Gap, because of The results of this study support Bretz et al. (1956) who higher flow velocities that kept icebergs moving through the believed the north flank of Rattlesnake Mountain displayed system. the greatest concentration of berg-rafted debris anywhere Iceberg-deposited erratics occur downstream of Wallula within the area impacted by outburst floods. This appears to Gap but are much less concentrated than on Rattlesnake apply to both erratics as well as bergmounds, which are rare- Mountain. In the backflooded Willamette Valley (south of ly reported outside the Pasco Basin (Allison 1933). More Portland, Oregon), for example, only about 1.9 erratics/km2 bergmounds and erratics accumulated here due to floodwa- were identified by Minervini et al. (2003), compared to al- ters that backflooded and pooled behind Wallula Gap, the most 38.6 erratics+bergmounds/km2 within the Rattlesnake first major constriction along the route of the floods beyond Mountain study area. However, it is acknowledged that the the Channeled Scabland. In the Pasco Basin, flow expansion concentration of erratics may be somewhat biased in favor and rising floodwaters behind Wallula Gap caused a tempo- of Rattlesnake Mountain because rafted debris is more vis- rary slowing of floodwaters (Bjornstad 2006). The slowing ible in the semi-arid, low shrub-steppe of eastern Washing- currents naturally directed icebergs to the quieter margins of ton, compared to the more humid and densely vegetated

E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 53 0 1 2 3 km Size Class >64 ft2 (>5.95 m2) 0 1 mi 32-64 (3.00-5.95) 16-32 (1.49-3.00) 8-16 (0.74-1.49) 4-8 (0.37-0.74) o 46 26' 2-4 (0.19-0.37) 1-2 (0.09-0.19) Above maximum flood (>380 m elev.) Rattlesnake Mountain Study Area

o 46 24'

o 46 22' R eal ts n a k eR di g

e

River

kima Firing RangYa N e Fig. 13: Distribution of erratics by size. Abb. 13: Verteilung von Findlingen gemäss Größe. o o o o 119 32' 119 30' 119 28' 119 26'

Willamette Valley. Although many erratics of the Willamette Erratic debris was observed up to an elevation of 362 m Valley could be buried within or beneath a cover of Quater- on Rattlesnake Mountain. Elsewhere in the Pasco Basin the nary sediments so too are many of the erratics on Rattle- author has observed erratics up to an elevation of 366 m on snake Mountain. Red Mountain (N46.301, W119.448). This height is close to As illustrated in Figures 7 and 9 the highest concentrations 372 m – the base elevation of a high flood-spillover chan- of erratic debris lies partway up the slopes of Rattlesnake nel along the east side of Wallula Gap (N46.0327, W118.9186) Mountain between 200- and 300 m elevation. The decrease that is incised to an elevation of 378 m. Thus, two independ- in erratic debris at lower elevations may be explained by in- ent lines of evidence appear to place the highest level for creased flow velocity that developed across the lower slopes outburst flooding in the Pasco Basin and upper Wallula Gap as the last of the floodwaters in Lake Lewis drained from the to at least 366 m and perhaps as high as 380 m shown in basin. Accordingly, the faster flows were sufficient enough Figures 2 and 3. Highest floodwater indicators downstream to prevent iceberg stagnation. Another possibility is more of Wallula Gap are only about 335 m elevation, therefore it erratics lie buried beneath a cover of Holocene-age eolian, appears that some hydraulic gradient still existed between slopewash or fluvial sediment at lower elevations. The de- the upper and lower ends of Wallula Gap during the larg- crease in erratic debris above 300 m is consistent with what est floods, although the flow was later impeded by ponding is known about the sizes of the many dozens or more out- that shifted downstream into the constricted reaches of the burst floods, only a limited number of which were extremely Columbia River Gorge (Benito & O’Connor 2003). These large (Benito & O’Connor 2003). The peak discharges for constrictions facilitated the high water levels observed in many of the floods were relatively small (<1 million 3m /sec) Wallula Gap – especially during waning stages of flooding. compared to the largest flood(s) estimated at 17.5 million m3/ sec (O’Connor & Baker 1992). Naturally, smaller outburst Size of Erratics floods would not pond as deep behind hydraulic constric- The size of erratics appears to be random with respect to tions at Wallula Gap and the Columbia River Gorge. elevation (Figure 13) (i.e., erratics of variable size exist up

54 E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 105 to 362 m elevation). The number of erratics, distributed over seven different size classes, is shown in Figures 10A and 14. The log-log trend of smaller erratics appears to drop off

104 around <0.75 m2, which may be due to sampling bias. Errat- ics larger than 0.7 m2 are more likely to extend above the low shrub-steppe vegetation rendering them more visible 103 and mappable. Smaller erratics, on the other hand, are less conspicuous and more likely to be overlooked, especially if located between survey transects. y = 152 x-1.6 7 102 ( ) In general, considerable amounts of erratic debris smaller 2 2 R = 0.99 than 0.09 m were observed to litter the surface, often mul- tiple clasts per square metre, especially below 300 m eleva-

Log Number of Erratics 10 tion within the study area. Even though it was not practical to map debris of this size, it might be possible to estimate the abundance of the <0.09 m2 size class based on the log- 1 log plot illustrated in Figure 14. Accordingly, the number of -2 -1 0 2 10 10 10 101 10 m2 larger erratics (0.7 m2 to 6 m2) appear to plot along a gener- Log Size ally straight-line, log-log function for the number of erratics (maximum cross-sectional area) versus size expressed as: N = 152 A–1.67 Fig. 14: Log-log plot of size vs. number of erratics measured on Rattlesnake Mountain. A linear log-log distribution may exist for erratics larger than where N = number of erratics, and A = maximum exposed 0.75 m2 in area. The straight-line equation and correlation coefficient2 (R = cross-sectional area in m2. 2 0.99) are based on the number of erratics mapped between the 0.75 m and Assuming erratics <0.75 m2 follow a similar straight-line 6 m2 size classes. log function, the actual number of erratics in the 0.1–0.2 m2) Abb. 14: Log-Log plot (Anmerkung: zweidimensionale grafische Darstellung size range might be on the order of several thousand, which von Daten auf Logarithmenbasis): Größe vs. Anzahl ausgemessener Find- linge auf Rattlesnake Mountain. Eine lineare Log-Log Verteilung existiert is considerably more than the 682 actually observed within möglicherweise für Findlinge größer als 0,75 2m im Untersuchungsgebiet. the study area (Figure 10A). Accordingly, the total number Die Geradengleichung und der Korrelationskoeffizient (R2 = 0,98) beruhen of even less-visible erratic clasts down to the next smaller auf der Zahl der Findlinge in den Größenklassen von 0,75 m2 sowie 6 m2. size class (0.01 m2) may exceed 300,000 based on the above equation. However, while it is possible the above equation

SW Rattlesnake ft m Mtn. 1500

400 NE 380 m ation 1000 300 300 m Elev

200

500

Fig. 15: A diagram to illustrate elevation-limited distribution of bergmounds on Rattlesnake Mountain. The largest icebergs were grounded on the gently sloping lake bottom, further from the ancient shores of short-lived Lake Lewis. Few well-developed bergmounds are observed above ~300 m elevation, even though floodwaters extended up to 380 m. Apparently, during the largest flood(s) all bergmound-producing icebergs became grounded below 75 vertical metres of maximum lake level because of their greater size. Abb. 15: Diagramm zur höhenbegrenzten Verteilung von Bergmounds auf Rattlesnake Mountain. Die größten Eisberge liefen weit entfernt von der alten Uferlinie von Lake Lewis auf der sanft abfallenden Seesohle auf Grund. Obwohl der See bis auf 380 m Seehöhe anstieg wurden nur wenige gut entwickelte Bergmounds oberhalb von 300 m Seehöhe gefunden, Während der größten Fluten liefen offenbar alle Bergmound erzeugenden Eisberge aufgrund ihrer Größe unterhalb von 75 Höhenmeter unter dem maximalen Seespiegel auf Grund.

E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 55 Fig. 16: Diagram showing variable flow and speed of floodwater across a portion of the study area. Where floodwaters were deeper to the right, eddy currents caused icebergs to collect and ground on upstream sides of gul- lies while in shallower more-restrictive waters (left side) icebergs tended to ground along the bottoms of gullies. Elevations in this standard USGS topographic map are in feet (conversion to metres = 0.3048). Abb. 16: Diagramm zur unterschiedlichen Ab- laufmenge und Geschwindigkeit von Flutwas- sers in einem Teil des Untersuchungsgebietes. Wo die Flutwasser auf der rechen Seite tiefer waren, verursachten foucaultsche Wirbelströ- me („eddy currents“), so dass sich Eisberge stromaufwärts an den Seiten der Ablaufrinnen stauten und auf Grund liefen, während in flacherem, eingegrenzterem Wasser (auf der linken Seite) die Eisberge eher auf dem Grund der Ablaufrinnen strandeten. Topographische Höhe in Fuß (Umrechnung in Meter = 0.3048). may not accurately reflect the true number of erratics for Plateau (Figure 1), and 2) the Colville Valley (Figures 1 and different size classes, the known bias introduced in size ver- 6). Thus, ice-rafted basalt may be derived from: 1) icebergs sus visibility suggests a log-log distribution may in fact rep- that calved off the Okanogan Lobe (and possibly the Colville resent a reasonable estimate. Lobe) into glacial Lake Columbia, or 2) a final flood associ- ated with the breakup of Lake Columbia several centuries Sources of Erratic Debris after the last Missoula floodAtwater ( 1987; Waitt 1994; The lithology of erratics is consistent with an origin from ar- Waitt et al. 2009). eas once in contact with the Cordilleran Ice Sheet. Argillite is closely associated with quartzite; both are meta-sedimentary Surficial Weathering and Roundness of Erratics rocks derived from the Proterozoic Belt Supergroup Orr( & The majority of erratics (especially granitic) on the surface Orr 1996). These rocks crop out along the eastern side of the show some degree of roundness and weathering (Figure 8). Purcell Trench at the site of the ice dam for glacial Lake Mis- Yet the spatial distribution of surface weathering and round- soula (Miller et al. 1999; Bjornstad & Kiver, 2012). Other ing appeared random with respect to elevation and location erratics include Cretaceous and Eocene plutonic rocks such within the study area. Because ice-rafted erratics undergo as granite and granodiorite and high-grade metamorphics little or no abrasion during ice transport, the rounding must (i.e., gneiss, schist), which crop out along the western side of have occurred either before or after melting out of icebergs. the Purcell Trench and in northern Washington (Stoffel et Like erratics on Rattlesnake Mountain, many glacial erratics al. 1991) (Figure 6). observed today near the former Cordilleran Ice Sheet show Some basalt, while indigenous to the area, was also en- some degree of rounding, either inherited from their place trapped in glacial ice and carried along with outburst flood- of origin prior to ice transport, or due to in-situ weathering waters. Although Columbia River basalt did not come in di- since the end of the Pleistocene. rect contact with the ice dam for Lake Missoula, there are Although it may be tempting to attribute weathering other places downstream where the ice sheet overrode ba- and roundness to in-situ weathering (via exfoliation and salt: 1) Okanogan lobe where it extended across Waterville spalling) since iceberg emplacement, this does not appear to

56 E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License be the case based on a limited number of cosmogenic-expo- the distribution of ice-rafted debris shown in Figure 7 seems sure ages on erratics (Keszthelyi et al. 2009). Three dated warranted. erratics shown in Table 1 are variably weathered (two of the Based on Archimedes’ principle the minimum size of an three strongly weathered), yet their exposure ages fall into a iceberg needed to float a boulder of granite composition (av- narrow range of time. All three erratics are associated with erage density ~2.70 g/cm3) is estimated at about 20 times the the last cycle (Late Wisconsin) glaciation between 16 and volume of the erratic boulder. In marine waters only the up- 17 thousand years. This amount of time appears insufficient per one-eighth of icebergs rise above water level (Bruneau to produce a strongly weathered erratic surface especially 2004). However, due to higher sediment loads and densities of considering some granitic erratics on Rattlesnake Mountain continental floodwaters, it is likely icebergs were even more show little or no weathering. Therefore, the degree of weath- buoyant and floated higher when compared to marine water ering and rounding of granitic erratics does not appear to be containing little or no suspended sediment. Nevertheless, the a reliable indicator of age since grounding of erratics and bulk of floating icebergs lie underwater and protrude well suggests at least some weathering and roundness of errat- below the waterline. Essentially, the larger the iceberg the ics is inherited from their place of origin, prior to being ice deeper it extended into temporary Lake Lewis (Figure 15). rafted to Rattlesnake Mountain. It follows then that larger icebergs would become grounded Because of the extremely limited number of dated errat- well away from the ancient shores of Lake Lewis. This would ics, however, the possibility still exists that other, undated er- explain why far fewer bergmounds lie above 300 m eleva- ratics could have undergone strong weathering in situ. Out- tion. Furthermore, the greater concentration of ice-rafted de- burst floods from previous glacial cycles going back to Early bris at lower elevations (Figure 7) is consistent with evidence Pleistocene (>780 ka) are known to have occurred in east- to indicate there were many more smaller floods versus large ern Washington (Patton and Baker 1978; Bjornstad et al. floods toward the end of the Wisconsin Glaciation (Waitt 2001; Pluhar et al. 2006). Erratics associated with these ear- 1980; Benito & O’Connor 2003). lier flood events would be expected to display strong weath- ering if continuously exposed at the ground surface. On the Flow Dynamics other hand, those erratics derived from pre-Wisconsin floods A notable pattern, illustrated in Figure 7, is the alignment that lie buried beneath an aggrading mantle of post-basalt of many erratics and erratic clusters along gullies. At lower Pleistocene sediments may have escaped surficial weather- elevations, erratics are concentrated along the south-facing ing. (upstream) slopes of gullies. At higher elevations, erratics lie mostly along the bases of these same gullies. The pattern of Origin of Bergmounds concentration for erratics along gullies is attributed to eddy Although it seems logical to ascribe bergmounds (Figure currents set up by floodwaters flowing at slightly different 4C) entirely to the buildup of ice-rafted debris from large, velocities across the uneven surface of Rattlesnake Moun- sediment-laden icebergs, the model for bergmound devel- tain. It is envisioned that, as floodwater levels rapidly low- opment may be more complicated. Not all topographic re- ered, icebergs were trapped in eddies that circulated in the lief observed on bergmounds is necessarily constructional. deep and slower moving water within the gullies. At higher Some of the relief may also be the result of increased erosion elevations where the lake was shallower, deep-rooted ice- around the flanks of the bergmound, which occurred during bergs migrated to the deeper water overlying the bottoms of or since flooding. This is suggested from trenching studies of gullies. At lower elevations icebergs were less constrained bergmounds, which reveal at least some of the interiors are because of deeper water and therefore more likely to become composed of mostly fine-grained sediments, which are not grounded on the high margins to the upstream side of gullies necessarily derived from ice-rafted debris (Fecht & Tall- where eddy circulation concentrated their movement (Fig- man 1978; Chamness 1993, 1994). Thus, coarser ice-rafted ure 16). debris may merely blanket bergmounds creating a lag-grav- Erratics are also more concentrated along roadways with- el cap that acts to armor and protect the underlying finer- in the study area; this concentration suggests exhumation grained sediments. Areas between bergmounds, lacking a during road construction from fine-textured, post-basalt coarse, armored cap, then were more susceptible to erosion sediments. Furthermore, cisterns dug by early 20th century by either: 1) receding floodwaters, or 2) post-flood eolian de- settlers in this older loose, fine sediment often produced er- flation. However, problematic is the almost perfectly sym- ratic boulders. These observations indicate: 1) many errat- metrical and conical shape of bergmounds that show little or ics from older floods lie buried within the post-basalt sedi- no sign of streamlining, which might be expected if subse- ment cover, and 2) there are far more erratic boulders on quently eroded by either moving floodwater or wind. Rattlesnake Mountain than those currently exposed at the surface. Some of these buried boulders could be associated Distribution of Ice-Rafted Debris with older, pre-Wisconsin (>120 ka) outburst floods (Patton Bretz et al. (1956) noted two populations of ice-rafted de- & Baker 1978; Bjornstad et al. 2001). bris along Rattlesnake Mountain. They reported bergmounds up to an elevation of 260 m; while single erratics or groups 7 Conclusions of erratics extended up to 335 m elevation they speculated the two populations were the result of two separate flood Ice-rafted debris accumulated in slackwater areas up to an events. However, knowing there were at least dozens (Waitt elevation of ~365 m within south-central Washington during 1980) to perhaps hundreds of separate flood events (Bjorns- repeated Pleistocene cataclysmic floods. Erosion by floods tad 2006) of various magnitudes, a different explanation for along the basin outlet at Wallula Gap indicates maximum

E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 57 flood depths approached 380 m. Floodwaters impounded be- References hind Wallula Gap temporarily created Lake Lewis and de- posited ice-rafted erratics and bergmounds along the gentle Allen, J.E., Burns, M. & Burns, S. (2009): Cataclysms on the Columbia. – slopes of Rattlesnake Mountain. Ice-rafted debris is of three 205 p.; Second Edition, Portland, Oregon (Ooligan Press). Allison, I.S. (1933): New version of the Spokane flood. – Bulletin of the types: 1) isolated erratics, 2) erratic clusters, and 3) berg- Geological Society of America, 44: 675–722. mounds. Unlike erratics and clusters, bergmounds display Allison, I.S. (1935): Glacial erratics in the Willamette Valley. – Bulletin of topographic relief in the form of scattered, low-relief, conical the Geological Society of America, 46: 605–632. mounds. Atwater, B.F. (1986): Pleistocene glacial-lake deposits of the Sanpoil River valley, northeastern Washington. – USGS Bulletin 1661, U.S. Geologi- The present study of ice-rafted debris was performed in a cal Survey. long-protected, sparsely vegetated, 60 km2 area on the north- Atwater, B.F. (1987): Status of glacial Lake Columbia during the last floods eastern flank of Rattlesnake Mountain. More than 2,100 lo- from glacial Lake Missoula. – Quaternary Research, 27: 182–201. cations of erratics >0.09 m2 and bergmounds were recorded Baker, V.R. (1978): Paleohydraulics and hydrodynamics of the Scabland Floods. – In: Baker, V.R., Nummedal, D. (eds.): The Channeled with a GPS unit. Additional information was gathered on Scabland. – Washington D.C. (National Aeronautics and Space Admin- 1) elevation, 2) lithology, 3) size, 4) roundness, 5) shape, and istration), 59–79. 6) surface characteristics of erratics. All but a few percent Baker, V.R. & Bunker, R.C. (1985): Cataclysmic Late Pleistocene flooding of erratics consist of rocks other than indigenous basalt; ap- from glacial Lake Missoula: A review. – Quaternary Science Reviews, 4: 1–41. proximately 75% are of granitic composition. Other litholo- Baker, V.R., Bjornstad, B.N., Busacca, A.J., Fecht, K.R., Kiver, E.P., gies, in order of decreasing abundance, are quartzite, diorite, Moody, U.L., Rigby, J.G., Stradling, D.F. & Tallman, A.M. (1991): argillite, basalt, gneiss, schist, and gabbro. Quaternary geology of the Columbia Plateau. – In: Morrison, R.B. 36Cl-exposure ages on three, widely dispersed erratics (ed.), Quaternary Nonglacial Geology, Conterminous U.S., DNAG [Decade of North American Geology], Geology of North America, were 16-17 ka, suggesting all were emplaced during the last Boulder, Colorado (Geological Society of America), v. K-2, 215–250. glacial cycle (i.e., Late Wisconsin). Most erratics are either Benito, G. & O’Connor, J.E. (2003): Number and size of last-glacial Mis- subrounded or rounded, followed by subangular; angu- soula floods in the Columbia River valley between Pasco Basin, Wash- lar clasts are least common. Greater than three quarters of ington, and Portland, Oregon. – Bulletin of the Geological Society of America, 115: 624–638. erratics are moderately to strongly weathered, but a wide Bjornstad, B. N., Fecht, K.R. & Pluhar, C.J. (2001): Long history of pre- range of weathering and roundness was observed on the Wisconsin, Ice-Age floods: Evidence from southeastern Washington cosmogenically-dated erratics suggesting these features are State. – Journal of Geology, 109: 695–713. likely inherited from their place of origin. Bjornstad, B.N. (2006): On the Trail of the Ice-Age Floods: Guide to the Mid-Columbia Basin. – 307 p.; Sandpoint, Idaho (Keokee). The distribution of ice-rafted debris is non-uniform; max- Bjornstad, B.N., & E.P. Kiver (2012): On the Trail of the Ice Age Floods: imum concentration is around 200 m elevation. Bergmounds The Northern Reaches. – 432 p.; Sandpoint, Idaho (Keokee). are mostly restricted to a band between 200- and 300 m el- Bjornstad, B.N., Babcock, R.S. & Last, G.V. (2007): Flood basalts and Ice evation; especially large icebergs that produced bergmounds age floods: Repeated late Cenozoic cataclysms of southeastern Wash- ington. – In: Stelling, P. & Tucker, D.S., (eds.), Floods, Faults and could not approach the shores of the ancient shoaling lake, Fire: Geological Field Trips in Washington State and Southwest British which extended up to 80 m higher. As floodwaters moved Columbia. – Geological Society of America Field Guide 9, 209–255, doi: across an uneven surface backwater eddies concentrated 10.1130/2007.fld009 (10). icebergs and their entrapped erratics along the upstream Bretz, J H. (1919): The late Pleistocene submergence in the Columbia valley of Oregon and Washington. – Journal of Geology, 27: 489–506. sides of northeast-trending gullies within the study area. Bretz, J H. (1923a): Glacial drainage on the Columbia Plateau. – Bulletin of This study affirms that the north slope of Rattlesnake Moun- the Geological Society of America, 34: 573–608. tain likely contains the most prolific erratic and bergmound Bretz, J H. (1923b): The Channeled Scabland of the Columbia Plateau. – population of anywhere along the route for glacial-outburst Journal of Geology, 31: 617–649. Bretz, J H. (1930): Valley deposits immediately west of the Channeled floods in the northwestern US. Scabland. – Journal of Geology, 38: 385–422. Bretz, J H., Smith, H.T.U. & Neff, G.E. (1956): Channeled Scabland of Acknowledgements Washington: New data and interpretations. – Bulletin of the Geological Society of America, 67: 957–1049. Bretz, J H. (1969): The Lake Missoula floods and the Channeled Scabland. – Special thanks go to Elysia Jennet, Bronwyn Ryan, and Rick Journal of Geology, 77: 505–543. Edwards who each spent a summer diligently collecting field Bruneau, S.E. (2004): Icebergs of Newfoundland and Labrador. – 64 p.; data. Jenna Gaston of the U.S. Fish and Wildlife Service as- St.Johns, Newfoundland, (Flanker Press). sisted with the special-use permit to access the restricted Chamness, M.A. (1993): An investigation of bergmounds as analogs to ero- sion control factors on protective barriers. PNL-8841, Pacific Northwest Fitzner/Eberhardt Arid Lands Ecology Reserve Unit of the Laboratory, Richland, Washington. Hanford Reach National Monument. Also appreciated is the Chamness, M.A. (1994): Bergmounds: A depositional or erosional feature? support of Gary Kleinknecht and Ivar Husa and helpful re- – Geological Society of America, Abstracts with Programs 26: A–307. views of earlier drafts of the manuscript by Jim O’Connor Denlinger, R.P. & O’Connell, D.R.H. (2010): Simulation of cataclysmic outburst floods from Pleistocene glacial Lake Missoula. – Bulletin of and Jim Knox. Reviews by colleagues at the Pacific North- the Geological Society of America, 122: 678–689. DOI: 10.1130/B26454.1. west National Laboratory, including Mickie Chamness, Fecht, K.R. & Tallman, A.M. (1978): Bergmounds along the western mar- George Last and Christopher Murray, further improved the gin of the Channeled Scablands, south-central Washington. – RHO- manuscript. Appreciated is an editorial review by Andrea BWI-SA-11, Rockwell Hanford Operations, Richland, Washington. Karlson, R.C. (2006): Investigation of Ice Age flood geomorphology and Currie and translations by Nik Foster. Vic Baker and Marek stratigraphy in Ginkgo Petrified Forest State Park, Washington: Im- Zreda provided 36Cl-exposure age dates for Rattlesnake plications for park interpretation. – Master of Science Thesis, Central Mountain erratics. Lastly, helpful reviews by Leszek Marks Washington University, Ellensburg, Washington, 146 p. and Jim Rose greatly improved the quality of the manuscript. Keszthelyi, L.P., Baker, V.R., Jaeger, W.L., Gaylord, D.R., Bjornstad, B.N., Greenbaum, N., Self, S., Thordarson, T., Porat, N. & Zreda,

58 E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License M.G. (2009): Floods of water and lava in the Columbia River Basin: onset of ice-age cataclysmic floods in eastern Washington during the Analogs for Mars. – In: O’Connor, J.E., Dorsey, R.J. & Madin, I.P., Early Pleistocene. – Quaternary Research, 65: 123–135. (eds.), Volcanoes to Vineyards: Geologic Field Trips Through the Dy- Shaw, J., Munro-Stasuik, M., Sawyer, B., Beaney, C., Lesemann, J.- namic Landscape of the Pacific Northwest. – Geological Society of E., Musacchio, A., Rains, B. & Young, R.R. (1999): The Channeled America Field Guide 15, 845–874, doi: 10.1130/2009.fld015 (34). Scabland: Back to Bretz. – Geology, 27: 605–608. Miller, F.K., Burmester, R.F., Miller, D.M., Powell, R.E. & Derkey, P.M. Smith, G.A. (1993): Missoula flood dynamics and magnitudes inferred from (1999): Digital geologic map of the Sandpoint 1- by 2-degree quadran- sedimentology of slack-water deposits on the Columbia Plateau. –Bul - gle, Washington, Idaho, and Montana. – Open-File Report 99–144, U.S. letin of the Geological Society of America, 195: 77–100. Geological Survey, Menlo Park, California. Stoffel, K.L., Joseph, N.L., Waggoner, S.Z., Gulick, C.W., Korosec, M.A. Minervini, J., O’Connor, J.E. & Wells, R.E. (2003): Inundation map, ice- & Bunning, B.B. (1991): Geologic Map of Washington – Northeast rafted erratics, and deposits of Late Pleistocene Missoula floods in the Quadrant. – Geologic Map GM-39, Washington Division of Geology Portland Basin and Willamette Valley, Oregon and Washington. – Geo- and Earth Resources, Olympia, Washington. logical Society of America, Abstracts with Programs, 34: 335. Waitt, R.B. (1980): About forty last-glacial Lake Missoula jökulhlaups O’Connor, J.E. (1993): Hydrology, hydraulics, and geomorphology of the through southern Washington. – Journal of Geology, 88: 653–679. Bonneville Flood. – GSA Special Paper 274, Geological Society of Waitt, R.B. (1985): Case for periodic, colossal jökulhlaups from Pleistocene America, Boulder, Colorado, 83 p. glacial Lake Missoula. – Bulletin of the Geological Society of America, O‘Connor, J.E. & Baker, V.R. (1992): Magnitudes and implications of peak 96: 1271–1286. discharges from glacial Lake Missoula. – Bulletin of the Geological So- Waitt, R.B. (1994): Scores of gigantic, successively smaller Lake Missoula ciety of America, 104: 267–279. floods through Channeled Scabland and Columbia Valley. – In: Swan- O’Connor, J.E. & Benito, G. (2009): Late Pleistocene Missoula floods – son D.A. & Haugerud, R.A. (eds.): Geologic Field Trips in the Pacif- 15,000–20,000 calendar years before present from radiocarbon dating. ic Northwest, Chapter 1K. – Geological Society of America, Boulder, – Geological Society of America, Abstracts with Programs, 41: 169. Colorado, pp. 1K-1 to 1K-88. Orr, E.L & Orr, W.N. (1996): Geology of the Pacific Northwest. – 409 p.; Waitt, R.B., Denlinger, R.P. & O’Connor, J.E. (2009): Many Monstrous New York (McGraw-Hill). Missoula floods down Channeled Scabland and Columbia Valley. – In: Patton, P.C. & Baker, V.R. (1978): New evidence of pre-Wisconsin flood- O’Connor, J.E, Dorsey, R.J., & Madin, I.P. (eds.): Volcanoes to Vine- ing in the channeled scabland of eastern Washington. – Geology, 6: yards: Geologic Field Trips through Dynamic Landscape of the Pacific 567–571. Northwest. – Geological Society of America Field Guide 15: 775–844, Pluhar, C.J., Bjornstad, B.N., Reidel, S.P., Coe, R.S. & Nelson, P.B. doi: 10.1130/2009.ftd015(33). (2006): Magnetostratigraphic evidence from the Cold Creek bar for

E&G / Vol. 63 / No. 1 / 2014 / 44–59 / DOI 10.3285/eg.63.1.03 / © Authors / Creative Commons Attribution License 59